Classification and identification of epileptic EEG signals based on signal enhancement

Abstract

Epilepsy is a common chronic brain disorder. The complexity and suddenness of seizures make it difficult to obtain EEG signals. Therefore, the detection and diagnosis of epilepsy is challenging. Aiming at the poor classification and recognition effect of small sample epileptic EEG signals, this paper proposes a classification and recognition method based on signal enhancement. Firstly, the improved sliding window weighting method is used to remove the noise component and enhance the target signal. Then discrete wavelet transform is used to select the appropriate frequency band information, and non-zero processing is performed on the selected frequency band information, which can achieve the effect of feature enhancement in the feature extraction stage. Finally, the extracted features are input into a support vector machine (SVM) classifier for classification and recognition. The experimental results on the public database show that the proposed method can effectively realize classification and recognition in the environment of small sample EEG signals, and the classification accuracy is 96.59%. This result proves the superiority of our method in the classification of healthy, inter-ictal state and seizure state EEG signals.